- VHDL 89.4%
- Verilog 9.1%
- MATLAB 1.3%
- Standard ML 0.2%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| Code | ||
| Documents | ||
| README.md | ||
FPGA-Implementation-Of-Adaptive-Noise-Cancellation
Speech is the most fundamental method of communication for humans, with a bandwidth of only 4 kHz. It can communicate information with the same emotion as a human voice. The voice signal has certain characteristics, such as being a one-dimensional signal with time as its independent variable. It is unpredictable, non-stationary, and the frequency spectrum does not remain consistent throughout time. Human speech contains substantial frequency components only up to 4 kHz while having an audible frequency range of 20 Hz to 20 kHz. The influence of interfering noise in the signals is the most prevalent issue in voice processing. The presence of background acoustic noise conceals the voice signal, reducing its intelligibility, and also affects speech communication in loud environments. Speech intelligibility is greatly reduced when background noise is present. Background noise is suppressed using noise reduction algorithms, which increase the perceived quality and intelligibility of speech. Due to the unpredictable nature of noise and the inherent complexity of speech, removing various forms of noise is challenging. In most noise reduction approaches, there is a trade-off between the quantity of noise removed and the speech distortions generated as a result of signal processing. In the field of voice enhancement, several approaches have been developed for this goal, including the spectral subtraction approach, wiener filter, and Kalman filter. The quality and intelligibility of the processed speech stream determine how well these approaches function. The goal of most methods is to enhance the voice signal to noise ratio.